الفهرس 
1. INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 172 |  |  |
| | | from | 172 | | |
Abstract
The real-time dynamic path planning and path tracking algorithm of a WMR using a single camera and distributed Networked Cameras is designed and implemented. The main contribution of this work is to solve some problems of path planning, mobile robot localization, and path following as follows;
1- Develop and implement an improved MSFM path planning algorithm, which is based on the robot dimensions (W, L) practically. The major goal is to generate a safe and optimized path. An image preprocessing dilation operation step, which based on the robot dimensions (W/2, L/2), is used to dilate the distance map that used with the MSFM to generate the desired path. As a result, the generated path is located as near as possible to the obstacle (s), by a sufficient distance determined by the robot dimensions, not in the middle between obstacles. Consequently, short travelling distances for the robot are expected to save the total traveling time and consume less power. The developed method is tested and applied practically with practical environments and painted maps.
2- The algorithm integrates the feature -based and the template-matching methods to segment and classify the environment into three images; the target point image, the obstacles image, and the robot image. The robot location is calculated from the robot image at each frame and there is a new development method to calculate the robot heading angle using template-matching algorithm was introduced in our work.
3- The path following controller uses the PD-like FLC in a decentralized form and compared with the conventional linear PD controller. The PD controller gains are tuned for certain reference path and certain linear and angular velocities. In the case of PD controller, the robot follows the path with
a considerable error. This issue is one of the disadvantages of the classic controllers for nonlinear systems. While the decentralized PD-like FLC successes to make the robot to follow the desired path. The decentralized form of the FLC has some advantages; simplicity of design which is needed only the error and the error change at its inputs, minimum calculation time and suitable for real time control applications, but it needs a large effort to build the rule bases for each stage to make a synchronization between them.
The overall system is experimentally verified and validated using a single camera and distributed networked cameras, under normal and perturbed conditions such as; light changes, pushing the robot away from the path, and changing the robot speed by loading and unloading the robot. The observed mean absolute error and the root mean square error in all cases are summarized in table 6.1. In the case of single camera, the overall algorithm used to grab images from the camera and to calculate the desired control signal is programmed and executed using only one PC. The algorithm is tested on Core i5 PC and Core2Due PC. The average execution time was observed is about 60 ms for the Core i5 PC and 80 ms for the latter, while the algorithm is programmed and executed using two PCs in the case of networked cameras. The first PC used to acquire and analyze images from cameras and the other PC used as a path following controller, the average execution time in this case is about 67 msec. The comparisons to other references are summarized at the end of chapter 5. The results show that the algorithm is robust to noise while following a safe and optimized path under different perturbed conditions.